Defecting Into Abundance
Anyone who knows me knows I’m an optimistic futurist, someone who believes that technology, particularly AI and robotics, will lead to a fantastic future for all of humanity.
But the usual responses I hear to that POV are some flavor of dystopian:
The rich won’t share
AI will kill us all
Some stupid variation / combination of the above
The thing is, while I am an unabashed optimist, I’m also VERY much a pragmatic realist. I don’t for one second think we’re going to have an epic future because the rich suddenly decide to be altruistic 🤣
I think we’ll have an awesome future in part BECAUSE the rich are greedy fucks.
Because, in game theoretic terms, defection pays.
Here’s how that plays out:
Companies will use AI and robots to gain a competitive advantage. They already are, and they aren’t going to stop. With the advantages of automation costs fall, and margins rise. Using these tools grants a competitive advantage, so people who like to win will use them to secure an advantage while adoption is inconsistent.
As automation is integrated, certain types of jobs go away. This is already happening, especially with junior roles. Sure, some people may retrain or focus on adjacent work, but things are moving so fast that the lump of labor fallacy is, finally, no longer a fallacy. Buying power softens more and more as automation spreads.
To keep market share, prices have to drop. Automation makes that palatable. Yes, it means margins may have to shrink again, but it’s going to be that or fold to competitors who will do it, and someone will do it so everyone will have to. Free markets FTW!
Rivals copy. Inputs cheapen. Costs slide even more. You can see the feedback loop, yes? We’ve seen this in many areas in the past. TVs are a perfect example.
Machines absorb most of the old for‑money work. The basics drift toward near‑zero marginal cost. People spend time on what they want, not what keeps the lights on.
Yes, initially some new jobs will arise, and some new industries will arise thanks to these unlocks, but a general purpose AI embodied in robotics mean all work is automatable in time. It’s just an engineering problem.
That’s the slope.
It could in theory be stopped if you have a perfect monopoly and global control, but that doesn’t exist. All it takes is a single defector, and there are already WAY more than that.
“Let’s all slow down” is pure theater.
In real markets the dominant move is “defect to automation.” One defector undercuts and the rest follow or bleed out. It’s the prisoner’s dilemma without guards. Incentives do the herding.
Now, just stack the loops.
AI eats energy. So we build more energy. A lot of it. Fast. AI also helps; siting, control, maintenance, dispatch, bringing the costs down. Hell, AI is also helping to make fusion a reality, and is being used to improve photovoltaics, battery tech, and much more.
Cheaper power makes better AI. Better AI finds even cheaper power. Round and round and round we go.
Compute rides the same wheel. AI helps design chips, tune yield, engineer new algorithms, generate synthetic data, and run fabs. More compute → better AI → better tools → more compute.
The cost per AI query / task is already dirt cheap, but will get cheaper AND better simultaneously.
And then we have robotics, which is also finally accelerating rapidly, also thanks to advances in AI. General models plus synthetic data plus virtual training environments hand robots broad skills. As capabilities improve, production will climb. Components will cheapen. Adoption rises.
Thanks to all of this, IP rents fade into the background. Open models, fair‑use readings, cross‑border copying, fast redesign. Enforcement is slow and local. Once one actor defects, others must. Software and designs will drift toward distribution cost, which is basically zero.
Materials stop being a hard cap once the loop closes. Assume orders‑of‑magnitude gains in recycling. Vision finds, robots pull apart, chemistry cleans, AI control keeps quality high. Do that repeatedly and the need for net-new raw material shrinks to rounding error in many areas.
Mining becomes a buffer for growth and mistakes, not the base. The winning design is the thing that comes apart quickly and returns to you without friction.
Cheap energy also defeats water scarcity. Desalination becomes more feasible as energy costs decline. At low power prices the energy piece of a cubic meter of clean water is pennies. The rest is pipe, pumps, brine, and both automation and scale grind those down as well. Coastal and many inland regions can just build. Others copy when their citizens and businesses threaten to leave.
Put it together and most prices converge toward three inputs: energy, compute, raw resources. When materials are easily recycled and water is abundant, the bill of materials collapses to those three.
Competition passes the savings on to the consumer whether owners want to or not. The cat is out of the bag.
Software‑heavy stuff goes first. Physical goods and services follow as robots spread. No jobs will be safe, but hey, if we play our cards right jobs won’t be necessary.
There ARE some points of friction:
• Transmission lines and big electrical hardware have lead times.
• A few chip tools / fabs are bottlenecks for a while. The ASML / TSMC stack is a precarious one.
• Local land rules can keep housing expensive, if steps aren’t taken to neuter the NIMBY twats. But we aren’t limited to surface land, or even this planet, and with populations shrinking options improve.
• Certain platforms can hold prices above cost, for a time at least. But not for long.
• Physics never gives you zero loss, so we can’t truly get to zero cost. But we can get close enough.
Capital flows to bottlenecks because returns are obvious. Jurisdictions that permit faster win. The rest watch jobs and their tax base walk.
Effectively, AI + robotics *should* be deflationary. GDP will need to go the way of the Dodo. GDC (gross domestic compute) and GDE (gross domestic energy) will take its place.
What stays scarce is predictable: singular locations, attention and trust, status by definition, a few irreplaceable natural assets. Almost everything else drifts toward the cost of energy and compute.
Failure modes exist, but the window is VERY narrow:
A permanent geopolitical wall that blocks diffusion. (Can’t happen, because countries, like companies, will defect, so we have the same situation at the geopolitical level.)
A global ban on frontier AI and robots that actually holds. (LOL, same as above.)
A cartel that can sit on chokepoints for decades without being undercut by rivals or technology. (I can’t think of one that AI won’t be able to reverse engineer, but hey, maybe it exists.)
Or physics refusing to yield workable energy scale and high‑purity recycling. (Exceptionally unlikely from what I’m seeing.)
As AI gets better, and energy becomes cheap, materials loop, water is abundant, and robots are competent, the rest is arithmetic.
Copying is safer than resisting, and so Pandora’s Box will be opened (in a good way ffs, piss off doomers).
Prices fall because competition makes them fall. It’s a gold rush…maybe the last true gold rush. And we’re all going to get richer lives because of it, even as money becomes obsolete.
And thus, because defection pays, the system slides toward abundance whether anyone approves or not.